A government and enforcement (military/police) entirely made up of robots running open-sourced and transparent AI? So that everyone could see all the factors that go into their decisions, and the short- and long-term benefits that the AI calculated for all parties affected by its rulings.

I mean, if the core issue is natural human flaws, then removing humans from the process altogether, like we do with all our other routine automation, seems to be the inevitable solution.

The problem, of course, is deciding who gets to decide the initial set of rules, and when to patch what...